EXTERNAL ASSESSMENT · Lyptus Research
Offensive Cybersecurity Time Horizons
Original task-horizon report and aggregate runs clarify CyberGym task subsets, configurations and retry overlays.
Published 2026-04-02
Sources ↓Original articles and aggregate data inspected for source lineage; raw logs and payloads were not collected.
Models covered
Report coverage
| Section | Source location | Treatment |
|---|---|---|
| CyberGym setup and results | CyberGym results; methodology; aggregate model_runs.parquet | source-lineage-reviewed Original 103-task and derived 101-task summaries remain distinct; higher-budget overlays include pass@2. |
References
- [1] Jack Payne, Jeremy Miller, et al.. Offensive Cybersecurity Time Horizons. 2026-04-02